Social learning can help individuals to efficiently acquire knowledge and skills. In the classroom, social learning often takes place in structured settings in which peers help, support, and tutor each other. Several protocols have been developed to make peer-assisted learning (PAL) more efficient. However, little attention has been devoted to how the transfer of knowledge is shaped by the social relationship between peers, and their relative positions in the social network. To address this gap, we combined social network analysis with an experimental social learning task, in which pupils (N = 135; aged 11-19) could use social information from their peers to improve their performance. We show that pupils' tendencies to use social information substantially decrease with the peer's distance in the social network. This effect is mediated by subjective closeness: pupils report feeling much closer to their friends than to their non-friends, and closeness strongly enhances social learning. Our results further show that, above and beyond these effects of network distance, social information use increases with the peer's social status (network centrality) and perceived smartness. Our results provide empirical evidence in a naturalistic setting for the role of specific network attributes in shaping pupils' willingness to learn from their peers. These findings illustrate the value of a social network approach for understanding knowledge transfer in the classroom and can be used to structure more effective peer learning. Impact and ImplicationsThis study shows how social network analyses can help teachers and practitioners boost learning outcomes. Social network data can inform matching of pupils for peer learning, and can help cultivating social cohesion in the classroom. Collecting social network data is a fast and cheap procedure, and may be easily incorporated in existing protocols aimed at promoting peer learning, knowledge transfer and positive school culture.

Introduction Analysis of covariance (ANCOVA) remains a widely misunderstood approach for dealing with group differences on potential covariates (Miller & Chapman, 2001). This misunderstanding of the ANCOVA has a long history and its discussion is dispersed across fields and journals, making it difficult to obtain a systematic overview. Here we present a network method to organize the results of a literature search conducted by 44 Master's students as part of the 2016 University of Amsterdam course "Good Research Practices". The ANCOVA Pitfall Dora wants to assess whether, in her own university, men earn more than women. She has access to the salaries of a subset of researchers, and, as expected, men earn significantly more than women (p < .005). But wait! The men in her sample are also older than the women, and this confounds the results: perhaps the salary difference is due to age rather than gender. To address this confound and "control for" age, Dora includes age as a covariate in an ANCOVA. This procedure is tempting but statistically problematic. The ANCOVA is easier to interpret correctly when age influences salary but does not differ across the groups. As explained in Miller and Chapman (2001; but see chapter 10 in Judd, McClelland, & Ryan, 2011, and Field, 2013, pp. 484-486), when groups differ on a covariate (e.g., age), removing the variance associated with the covariate also removes the shared variance associated with the group (e.g., gender). As a result, the grouping variable loses some of its representativeness. This occurs mostly when groups are pre-existing and are not obtained by random assignment (Jamieson, 2004). As an example, assume one has access to the height of several mountain peaks in the Himalayas and the Pyrenees (Cohen & Cohen, 1983). One may test whether the mountain ranges differ in height and it may be tempting to include air pressure as a covariate; after all, air pressure differs across the

Adolescence is a period of emerging independence, in which adolescents face difficult decisions, including those that involve risk for health and well-being. Previous research suggests that learning from others might be a prominent strategy of adolescents to inform these difficult decisions. However, there is a gap in the literature that addresses the active role adolescents may have in gaining information about others' behavior (i.e., social information search). Here, we investigated when and how much social information adolescents search before making decisions under risk and ambiguity, using a novel social search paradigm. In this paradigm, adolescents were able to reveal real information about their classmates' choices before deciding on their final choice. Our two experiments suggest that social information search can be broken down in two independent decisions: first the decision to initiate search, followed by the decision to continue search. Several factors motivated initiation of search, including: (a) the difficulty of the choice, (b) uncertainty about the outcome, and (c) the magnitude of the reward at stake. Search took generally longer when adolescents faced information not in line with their initial preference. Finally, we observed that adolescents used the sampled social information to inform their risky-choice behavior. Taken together, these results provide novel insights into the dynamics of peer influence in adolescence and stress the importance of treating adolescents not only as receivers, but as active agents searching for social information.